Step 4: Assign System Security Category

Once the security impact levels have been selected, reviewed and adjusted as necessary for the security objectives of each individual information type processed by an information system, it is necessary to assign a system security category based on the aggregate of information types. The Step 4 activities include the following:

(i) review identified security categorizations for the aggregate of information types;

(ii) determine the system security categorization by identifying the high water mark for each of the security objectives (confidentiality, integrity, availability) based on the aggregate of the information types;

(iii) adjust the high water mark for each system security objective, as necessary, by applying the factors discussed below;

(iv) assign the overall information system impact level based on the highest impact level for the system security objectives; and

(v) document all security categorization determinations and decisions.

Since the impact values (i.e., levels) for confidentiality, integrity, and availability may not always be the same for a particular information system, the high water mark concept is used to information system will generally be the highest impact level for the security objectives (confidentiality, integrity, and availability) associated with the aggregate of system information types. Thus, a low-impact system is defined as an information system in which all three of the security objectives are low. A moderate-impact system is an information system in which at least one of the security objectives is moderate and no security objective is greater than moderate. And finally, a high-impact system is an information system in which at least one security objective is high.

Guidelines for System Categorization

In some cases, the impact level for a system security category will be higher than any security objective impact level for any information type processed by the system.
The primary factors that most commonly raise the impact levels of the system security category above that of its constituent information types are aggregation and critical system functionality. Additionally, variations in sensitivity/criticality with respect to time may need to be factored into the impact assignment process. Some information loses its sensitivity in time (e.g., economic/commodity projections after they’ve been published). Other information is particularly critical at some point in time (e.g., weather data in the terminal approach area during aircraft landing operations). This section provides some general guidelines regarding how aggregation, critical functionality, and other system factors may affect system security categorization.

In order to effectively accomplish this step, various stakeholders (e.g., management, operational personnel, or security experts) may need to be involved in decisions regarding system-level impact assessments. The following factors should be considered in adjusting the system security objective impact levels.

Some information may have little or no sensitivity in isolation but may be highly sensitive in aggregation. In some cases, aggregation of large quantities of a single information type can reveal sensitive patterns and plans, or facilitate access to sensitive or critical systems. In other cases, aggregation of information of several different and seemingly innocuous types can have similar effects. In general, the sensitivity of a given data element is likely to be greater in context than in isolation (e.g., association of an account number with the identity of an individual and/or institution). The availability, routine operational employment, and sophistication of data aggregation and inference tools are all increasing rapidly. If review reveals increased sensitivity or criticality associated with information aggregates, then the system security objective impact levels may need to be adjusted to a higher level than would be indicated by the security impact levels associated with any individual information type. This could be implemented by incorporating a statement that explains the aggregation and potential security objective affected as well as the modification to impact levels.

Compromise of some information types may have low impact in the context of a system’s primary function but may have much more significance when viewed in the context of the potential impact of compromising:

  • Other systems to which the system in question is connected, or
  • Other systems which are dependent on that system’s information.

Access control information for a system that processes only low impact information might initially be thought to have only low impact security objectives. However, if access to that system might result in some form of access to other systems (e.g., over a network), the sensitivity and criticality attributes of all systems to which such indirect access can result needs to be considered. Similarly, some information may, in general, have low sensitivity and/or criticality security objectives. However, that information may be used by other systems to enable extremely sensitive or critical functions (e.g., air traffic control use of weather information or use of commercial flight information to identify military combat transport systems). Loss of data integrity, availability, temporal context, or other context can have catastrophic consequences.

There are times when a system security objective impact level should be elevated based on reasons other than its information.  An elevation based on extenuating circumstances can be more apparent by comparing the original security categorization to the business impact analysis. If the system was categorized based on FIPS 199 at a Moderate overall impact level but the system owner has determined it needs to be operational within 4-8 hours of a disruption irrespective of the aggregated information type availability security impact level assigned, then there is a disconnect that might be caused by the system’s extenuating circumstances. Agencies must customize the information system availability security impact level as appropriate to obtain full value and accuracy.

Public Information Integrity

Most Federal agencies maintain web pages that are accessible to the public. The vast majority of these public web pages permit interaction between the site and the public. In some cases, the site provides only information. In other cases, forms may be submitted via the website (e.g., applications for service or job applications). In some cases, the site is a medium for business transactions. Unauthorized modification or destruction of information affecting external communications (e.g., web pages, electronic mail) may adversely affect operations and/or public confidence in the agency. In most cases, the damage can be corrected within a relatively short period of time, and the damage is limited (impact level is low). In other cases (e.g., very large fraudulent transactions or modification of a web page belonging to an intelligence/security community component), the damage to mission function and/or public confidence in the agency can be serious. In such cases, the integrity impact associated with unauthorized modification or destruction of a public web page would be at least moderate.

Catastrophic Loss of System Availability

Either physical or logical destruction of major assets can result in very large expenditures to restore the assets and/or long periods of time for recovery. Permanent loss/unavailability of information system capabilities can seriously hamper agency operations and, where direct services to the public are involved, have a severe adverse effect on public confidence in Federal agencies. Particularly in the case of large systems, FIPS 199 criteria suggest that catastrophic loss of system availability may result in a high availability impact level. Whether or not the impact level of system availability should be high (and subsequent high system security impact level) is dependent on other factors, such as cost and criticality of the system, rather than on the security impact levels for the information types being processed by the system.

Large Supporting and Interconnecting Systems

Large or complex information systems composed of multiple lower level systems often require additional consideration regarding assignment of system security categorization. This section will provide guidelines for applying and interrelating individual system security categorization results to enterprise organizations, large supporting infrastructures (such as general support systems, data warehouse applications, large data storage units, server farms, and information repositories), and interconnecting systems.

Upon security categorization identification for all information systems interacting with large infrastructure systems, senior IT and security officials have possession of valuable information that can now enable an enterprise wide security perspective. One significant activity includes levying an overall security categorization for the agency’s supporting network infrastructures. Since networks, as well as other general support systems, do not inherently “own” mission-based or management and support information types, the infrastructure’s categorization is based on the aggregation of the information systems’ security categorizations. In other words, the infrastructure’s security categorization is the high water mark of the supported information systems and is based on the information types processed, flowed, or stored on the network or general support system. Together, the top down enterprise wide threat assessment and bottom up security assessment derived by aggregation will allow an organization to look at its risk profile from a comprehensive and balanced view. Further, this analysis will ensure the proper application of common security controls supporting the multiple information systems and the protection provided by those controls are inherited by the individual systems.

Critical Infrastructures and Key Resources

Where the mission served by an information system, or the information that the system processes, affects the security of critical infrastructures and key resources, the harm that results from a compromise requires particularly close attention. In this case, an effect on security might include a significant reduction in the effectiveness of physical or cyber security protection mechanisms, or facilitation of a terrorist attack on critical infrastructures and key resources. Accordingly, the system security categorization should be carefully determined when a loss of confidentiality, integrity, or availability will result in a negative impact on the critical infrastructures and key resources.

The Critical Information Infrastructure Act of 2002, Public Law 107-296 §§ 211-215 of November 25, 2002 (codified as 6 U.S.C. 131-134), defines the term “critical infrastructure information” to mean information not customarily in the public domain and related to the security of critical infrastructure or protected systems. Should information types be aligned with Critical Infrastructures, then action should be taken to ensure compliance with Homeland Security Presidential Directive No. 7 (HSPD 7) and to initiate an interdependency analysis.

Privacy Information

The E-Government Act of 2002 complements privacy protection requirements of the Privacy Act of 1974. Under the terms of these public laws, Federal government agencies have specific responsibilities regarding collection, dissemination or disclosure of information regarding individuals.

The September 26, 2003 OMB Memorandum M-03-22, “OMB Guidance for Implementing the Privacy Provisions of the E-Government Act of 2002,” puts the privacy provisions of the E-Government Act of 2002 into effect. The guidance applies to information that identifies individuals in a recognizable form, including name, address, telephone number, Social Security Number, and e-mail addresses. OMB instructed agency heads “to describe how the government handles information that individuals provide electronically, so that the American public has assurances that personal information is protected.” Under these public laws and executive policies, it is necessary to broaden the definition of “unauthorized disclosure” to encompass any access, use, disclosure, or sharing of privacy-protected information among Federal government agencies when such actions are prohibited by privacy laws and policies. Since most privacy regulations focus on access, use, disclosure, or sharing of information, privacy considerations are dealt with in this guideline as special factors affecting the confidentiality impact level. In establishing confidentiality impact levels for each information type, responsible parties must consider the consequences of unauthorized disclosure of privacy information (with respect to violations of Federal policy and/or law).

Agencies are required to conduct Privacy Impact Assessments (PIAs) before developing IT systems that contain personally identifiable information or before collecting personally identifiable information electronically. The impact of privacy violations should consider any adverse effects experienced by individuals or organizations as a result of the loss of PII confidentiality. Examples of adverse effects experienced by individuals may include blackmail, identity theft, discrimination, or emotional distress. Examples of adverse effects experienced by organizations may include administrative burden, financial losses, loss of public reputation and confidence, and the penalties associated with violation of the relevant statutes and policies.

Categorizations should be reviewed to ensure that the adverse effects of a loss of PII confidentiality have been adequately factored into impact determinations. The confidentiality impact level should generally fall into the moderate range.

Trade Secrets

There are several laws that specifically prohibit unauthorized disclosure of trade secrets (e.g., 7 U.S.C., Chapter 6, Subchapter II, Section 136h and 42 U.S.C., Chapter 6A, Subchapter XII, Part E, Section 300j-4(d)(1)). Systems that store, communicate, or process trade secrets will generally be assigned at least a moderate confidentiality impact level.